gfp-ring-detector

Analyzes granular level fluorescent microscopy images and counts the amount of granules with and without rings on their membrane.

Made for use at ELTE Anatómiai Sejt- és Fejlődésbiológiai Tanszék, especially in context of the role of ubiquitin in cell autophagy.

Dependencies

See conda-env.yml.

order.sh uses symbolic links, so linux is recommended.

Tested using zsh.

Notable executables

pipeline.py

Main product.

Analyzes an image and saves the result or (in interactive mode) displays it using the 5D Viewer.

• TODO Exit as soon as 5D Viewer is closed
• TODO Open 5D Viewer early and push new images in realtime
• TODO Measure and estimate computing time of each step
• TODO Do not depend on NanoImagingPack

Arguments

1. fname_template [semi-mandatory]
   Filename of all channel images. Use {} to denote the channel number. Example: python pipeline.py "képek/MyImage_z1_c{}.tif"
2. folder [optional]
   Output folder. If not supplied, interactive mode is activated. If supplied empty string (python pipeline.py "képek/MyImage/MyImage_c{}.tif" ""), the parent folder of the channel images is used.

process_all.sh

Walks the dataset in the képek/ folder looking for folders named *_Files, and analyzes the *_c<channel num>.tif files (channel num should be 0, 1 and 2) in it using pipeline.py.

• TODO Estimate finishing time
• TODO Communicate between parallel instances
• TODO Ability to spawn sub-processes automatically

Arguments

1. NUM_INSTANCES [optional]
   Number of threads this script starts or was started manually. If you'd like to make the script start multiple threads of itself, omit INSTANCE_ID
2. INSTANCE_ID [optional]
   An unique number between 1 and NUM_INSTANCES. The script will ignore all except every NUM_INSTANCESth file, this argument specifies which to process.

order.sh

Walks the képek/ folder, and if it finds a stats.txt named file, links the folder of the file (and the composite.tif file from it) to the newly-created ordered/ folder.

Environment variables

• $NO_RM [optional]
  To prevent emptying the ordered/ folder, set the $NO_RM env var to a non-empty value.

• $ZIP_FILE [optional] Creates a file with this name the most notable files from the ordered/ folder. Compatible with $NO_RM

• TODO $STREAMING environment variable not really works, fix and document it

Preconditions

• [optional] Run generate-composites.sh once for the dataset or you'll get some broken links
• Run process_all.sh in order to initialize and/or refresh the statistics for each image

generate-composites.sh

Walks the képek/ folder, and if it finds any relevant microscopy image, creates one named composite.tif and one named composite.jpg.

Useful for order.sh

Other tools

import_multipoint.ijm

ImageJ macro for importing generated coordinate lists

process_feh.sh

Useful if feh ordered/ command was issued. Finds the open feh window's title, extracts the folder, and analyzes the images in it using pipeline.py interactively.

• TODO React to a hotkey in the background

find-same.py

Locates the original raw files of steps.jpg in the local képek/ folder

find-centers.py

Performs grayscale Hough transformation on the input image.

Note: We are cheating, it's just a convolution with a ring-shaped kernel, but more precise

update.sh

Downloads the dataset from OneDrive if everything is properly configured

Arguments

1. zvi_file
   Input image. The 3rd zoom is considered from it.

5d.py

Opens a 4D ZVI file in 5D Viewer

Arguments:

1. zvi_file
   The Zeiss Microscopy Image to open in 5D Viewer

Demo

v1.x

Here is a low resolution demonstration of the internals of the first iteration, from 2021H2:

v2.x

Reimagined version with true hough transformation, better quality code, and integrated 5D Viewer built on top of NanoImagingPack written from scratch.

License

Not determined yet. Please ask the author (Domonkos Lezsák) if you are planning to use this software.

If the author is unavailable for three months, consider this repository to be licensed under CC-NC-BY-SA.